Pub Date : 2022-05-16DOI: 10.1109/CITDS54976.2022.9914138
M. Nsaif, Gergely Kovásznai, Mohammed G. K. Abboosh, Ali Malik, R. Fréin
Traffic classification is a crucial aspect for Software-Defined Networking functionalities. This paper is a part of an on-going project aiming at optimizing power consumption in the environment of software-defined datacenter networks. We have developed a novel routing strategy that can blindly balance between the power consumption and the quality of service for the incoming traffic flows. In this paper, we demonstrate how to classify the network traffic flows so that the quality of service of each flow-class can be guaranteed efficiently. This is achieved by creating a dataset that encompasses different types of network traffic such as video, VoIP, game and ICMP. The performance of a number of Machine Learning techniques is compared and the results are reported. As part of future work, we will incorporate classification into the power consumption model towards achieving further advances in this research area.
{"title":"ML-Based Online Traffic Classification for SDNs","authors":"M. Nsaif, Gergely Kovásznai, Mohammed G. K. Abboosh, Ali Malik, R. Fréin","doi":"10.1109/CITDS54976.2022.9914138","DOIUrl":"https://doi.org/10.1109/CITDS54976.2022.9914138","url":null,"abstract":"Traffic classification is a crucial aspect for Software-Defined Networking functionalities. This paper is a part of an on-going project aiming at optimizing power consumption in the environment of software-defined datacenter networks. We have developed a novel routing strategy that can blindly balance between the power consumption and the quality of service for the incoming traffic flows. In this paper, we demonstrate how to classify the network traffic flows so that the quality of service of each flow-class can be guaranteed efficiently. This is achieved by creating a dataset that encompasses different types of network traffic such as video, VoIP, game and ICMP. The performance of a number of Machine Learning techniques is compared and the results are reported. As part of future work, we will incorporate classification into the power consumption model towards achieving further advances in this research area.","PeriodicalId":271992,"journal":{"name":"2022 IEEE 2nd Conference on Information Technology and Data Science (CITDS)","volume":"1779 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129567743","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-16DOI: 10.1109/CITDS54976.2022.9914377
Hwawon Hwang, Yojin Kim, Seunghye Lee, Heejeong Choi, Pilsung Kang, Yongha In, Wonwoo Ro, Namwook Kang
Petrochemical companies put much effort into maximizing productivity and optimizing TCO(Total Cost of Operation) by reducing the unplanned downtime for stable operation of assets since unplanned downtime of assets leads to colossal production loss and environmental safety accidents. The PdM (Predictive Maintenance) solution is required to predict prognostic abnormal behavior of assets before the time when asset fault occurs, give warning alarm to engineers, and help them take proactive measures by diagnosing the fault cause and guiding suitable measures.In this research, the PdM model has been developed using Variational AutoEncoder and Isolation Forest algorithms to detect the prognostic abnormal behavior of assets before the unplanned shutdown. Moreover, PdM model for diagnosing the possible causes of abnormal behavior of the centrifugal compressor has also been developed to help domain field engineers take the suitable measures before the unplanned shutdown of the asset. By applying the PdM model to actual data of centrifugal compressor in petrochemical process, the PdM model has been successfully validated and shown feasible results.
{"title":"Development of Machine Learning based Model for Anomaly Detection and Fault Cause Diagnosis of Assets in Petrochemical Industries","authors":"Hwawon Hwang, Yojin Kim, Seunghye Lee, Heejeong Choi, Pilsung Kang, Yongha In, Wonwoo Ro, Namwook Kang","doi":"10.1109/CITDS54976.2022.9914377","DOIUrl":"https://doi.org/10.1109/CITDS54976.2022.9914377","url":null,"abstract":"Petrochemical companies put much effort into maximizing productivity and optimizing TCO(Total Cost of Operation) by reducing the unplanned downtime for stable operation of assets since unplanned downtime of assets leads to colossal production loss and environmental safety accidents. The PdM (Predictive Maintenance) solution is required to predict prognostic abnormal behavior of assets before the time when asset fault occurs, give warning alarm to engineers, and help them take proactive measures by diagnosing the fault cause and guiding suitable measures.In this research, the PdM model has been developed using Variational AutoEncoder and Isolation Forest algorithms to detect the prognostic abnormal behavior of assets before the unplanned shutdown. Moreover, PdM model for diagnosing the possible causes of abnormal behavior of the centrifugal compressor has also been developed to help domain field engineers take the suitable measures before the unplanned shutdown of the asset. By applying the PdM model to actual data of centrifugal compressor in petrochemical process, the PdM model has been successfully validated and shown feasible results.","PeriodicalId":271992,"journal":{"name":"2022 IEEE 2nd Conference on Information Technology and Data Science (CITDS)","volume":"209 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134097652","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-16DOI: 10.1109/CITDS54976.2022.9914171
Ferdous Zeaul Islam, Rifat Islam, Ashfaq Jamil, S. Momen
Rainfall is a crucial weather parameter in the context of Bangladesh. Prediction of rainfall can effectively aid the decision making process for agriculture and natural disaster management of the country. However the chaotic nature of rainfall due to climate change has made the task of rainfall prediction challenging through traditional statistical models. In this study, we analyze the performance of six machine learning algorithms: Decision Tree (DT), K-Nearest Neighbours (KNN), Random Forest (RF), Extreme Gradient Boosting (XGB), Light Gradient Boosting (LGB) and Multi-Layered Perceptron (MLP) in predicting daily rainfall as both regression and classification. In addition we try out an approach called Zero Inflated Regression (ZIR) to address the excessive amount of zero rainfall values in the dataset. The models were trained with and without feature selection and/or sampling techniques (for classification). During training 10-fold cross validation and hyperparameter tuning was performed on the train set and afterwards the selected models were applied to the test set for evaluation. For regression LGB with SelectKBest feature selection had the best performance on the test set with R2-score of 0.203, MAE of 6.40 and RMSE of 15.44. Among the classifiers, XGB with no feature selection and no sampling technique resulted with best test accuracy of 0.787 and test macro fl-score of 0.62. The ZIR model consisting of XGB classifier and LGB regressor with no feature selection yielded R2-score of 0.189, MAE of 5.789 and RMSE of 15.575 on the test set. Interestingly the ZIR models produced lower MAE compared to the regression models but the regression models had better R2-score.
{"title":"Evaluation of Machine Learning Methods for Predicting Rainfall in Bangladesh","authors":"Ferdous Zeaul Islam, Rifat Islam, Ashfaq Jamil, S. Momen","doi":"10.1109/CITDS54976.2022.9914171","DOIUrl":"https://doi.org/10.1109/CITDS54976.2022.9914171","url":null,"abstract":"Rainfall is a crucial weather parameter in the context of Bangladesh. Prediction of rainfall can effectively aid the decision making process for agriculture and natural disaster management of the country. However the chaotic nature of rainfall due to climate change has made the task of rainfall prediction challenging through traditional statistical models. In this study, we analyze the performance of six machine learning algorithms: Decision Tree (DT), K-Nearest Neighbours (KNN), Random Forest (RF), Extreme Gradient Boosting (XGB), Light Gradient Boosting (LGB) and Multi-Layered Perceptron (MLP) in predicting daily rainfall as both regression and classification. In addition we try out an approach called Zero Inflated Regression (ZIR) to address the excessive amount of zero rainfall values in the dataset. The models were trained with and without feature selection and/or sampling techniques (for classification). During training 10-fold cross validation and hyperparameter tuning was performed on the train set and afterwards the selected models were applied to the test set for evaluation. For regression LGB with SelectKBest feature selection had the best performance on the test set with R2-score of 0.203, MAE of 6.40 and RMSE of 15.44. Among the classifiers, XGB with no feature selection and no sampling technique resulted with best test accuracy of 0.787 and test macro fl-score of 0.62. The ZIR model consisting of XGB classifier and LGB regressor with no feature selection yielded R2-score of 0.189, MAE of 5.789 and RMSE of 15.575 on the test set. Interestingly the ZIR models produced lower MAE compared to the regression models but the regression models had better R2-score.","PeriodicalId":271992,"journal":{"name":"2022 IEEE 2nd Conference on Information Technology and Data Science (CITDS)","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132883757","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-16DOI: 10.1109/CITDS54976.2022.9914284
R. Trencsényi, L. Czap
The center of attraction of the present study is the application of neural networks for combining data arising from dynamic audiovisual sources made by ultrasound and magnetic resonance imaging methods, which store image and sound signals recorded during human speech. The objectives of machine learning are tongue contours fitted to the frames of the audiovisual packages by automatic contour tracking algorithms.
{"title":"Articulatory Data of Audiovisual Records of Speech Connected by Machine Learning","authors":"R. Trencsényi, L. Czap","doi":"10.1109/CITDS54976.2022.9914284","DOIUrl":"https://doi.org/10.1109/CITDS54976.2022.9914284","url":null,"abstract":"The center of attraction of the present study is the application of neural networks for combining data arising from dynamic audiovisual sources made by ultrasound and magnetic resonance imaging methods, which store image and sound signals recorded during human speech. The objectives of machine learning are tongue contours fitted to the frames of the audiovisual packages by automatic contour tracking algorithms.","PeriodicalId":271992,"journal":{"name":"2022 IEEE 2nd Conference on Information Technology and Data Science (CITDS)","volume":"35 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131287088","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-16DOI: 10.1109/CITDS54976.2022.9914219
Sonia Perez-Gamboa, Qingquan Sun, Amir Ghasemkhani
Sensor based human activity recognition has obtained more attentions due to its low-cost, low-data throughput, and immunity to environmental effects. However, traditional work in this field mainly focuses on the recognition of simple and small volume human activities. This work targets complicated, correlated and larger size of human activity recognition. In this paper, a multi-layer hybrid deep learning model is built with convolutional neural networks (CNN) and long short-term memory (LSTM). The multi-layer architecture improves the learning and exploration capacity of local features and temporal dependencies, and the hybrid architecture enriches the diversity for data fusion. In addition, Bayesian optimization is applied to the hybrid model to get the optimal parameters and best performance. The experimental results demonstrate the effectiveness of the proposed model with a recognition rate of 89% for 27 correlated activities. Its performance is better than traditional machine learning and other hybrid deep learning models.
{"title":"A Wireless Sensor Based Multi-layer Hybrid Deep Learning Model for Highly Correlated Human Activity Recognition","authors":"Sonia Perez-Gamboa, Qingquan Sun, Amir Ghasemkhani","doi":"10.1109/CITDS54976.2022.9914219","DOIUrl":"https://doi.org/10.1109/CITDS54976.2022.9914219","url":null,"abstract":"Sensor based human activity recognition has obtained more attentions due to its low-cost, low-data throughput, and immunity to environmental effects. However, traditional work in this field mainly focuses on the recognition of simple and small volume human activities. This work targets complicated, correlated and larger size of human activity recognition. In this paper, a multi-layer hybrid deep learning model is built with convolutional neural networks (CNN) and long short-term memory (LSTM). The multi-layer architecture improves the learning and exploration capacity of local features and temporal dependencies, and the hybrid architecture enriches the diversity for data fusion. In addition, Bayesian optimization is applied to the hybrid model to get the optimal parameters and best performance. The experimental results demonstrate the effectiveness of the proposed model with a recognition rate of 89% for 27 correlated activities. Its performance is better than traditional machine learning and other hybrid deep learning models.","PeriodicalId":271992,"journal":{"name":"2022 IEEE 2nd Conference on Information Technology and Data Science (CITDS)","volume":"117 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123309910","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-16DOI: 10.1109/CITDS54976.2022.9914145
L. Buttyán, Roland Nagy, Dorottya Papp
The Internet of Things is quickly developing and it enables exciting new applications, but at the same time, it also brings new security risks. In particular, embedded IoT devices may be subject to malware infection, undermining the trustworthiness of IoT systems. Malware detection on IoT devices is challenging due to their resource constraints, and antivirus tools developed for desktop PCs and servers are not directly applicable for them. In an earlier paper, we proposed a lightweight antivirus solution for IoT devices, called SIMBIoTA. In this paper, we propose SIMBIoTA++, an improvement on SIMBIoTA in terms of resource requirements. We also present a graph theory and measurement-based argument for selecting an appropriate similarity threshold, which is a key parameter in both SIMBIoTA and SIMBIoTA++.
{"title":"SIMBIoTA++: Improved Similarity-based IoT Malware Detection","authors":"L. Buttyán, Roland Nagy, Dorottya Papp","doi":"10.1109/CITDS54976.2022.9914145","DOIUrl":"https://doi.org/10.1109/CITDS54976.2022.9914145","url":null,"abstract":"The Internet of Things is quickly developing and it enables exciting new applications, but at the same time, it also brings new security risks. In particular, embedded IoT devices may be subject to malware infection, undermining the trustworthiness of IoT systems. Malware detection on IoT devices is challenging due to their resource constraints, and antivirus tools developed for desktop PCs and servers are not directly applicable for them. In an earlier paper, we proposed a lightweight antivirus solution for IoT devices, called SIMBIoTA. In this paper, we propose SIMBIoTA++, an improvement on SIMBIoTA in terms of resource requirements. We also present a graph theory and measurement-based argument for selecting an appropriate similarity threshold, which is a key parameter in both SIMBIoTA and SIMBIoTA++.","PeriodicalId":271992,"journal":{"name":"2022 IEEE 2nd Conference on Information Technology and Data Science (CITDS)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121020323","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-16DOI: 10.1109/CITDS54976.2022.9914149
Tahmina Khanom Tandra, Fehima Tajrian, Afia Hossain, M. T. Kawser, Mohammad Rubbyat Akram, A. Shams
To broaden user experience and support a wide range of bandwidth hungry devices cellular operators are adopting 5G network. However, the predominance of Inter-Cell Interference (ICI) in 5G stands as a hurdle for cell edge UEs. With increase in UE velocity, the effect of Doppler shift becomes more prominent, resulting in a significant drop in cell edge and mean data rates. A potential solution to improve the network service quality in the cell edge and reduce the impact of ICI for mobile users is to provide the UE with the best signal quality through coordination among multiple eNodeBs (eNB) located in different cell sites i.e., by virtually forming a massive antenna array with the coordinated eNB, a technique popularly known as Joint Transmission Coordinated Multipoint (JT CoMP). This paper investigates the performance of JT CoMP based heterogeneous network (HetNet) for UEs at different velocities while closed loop spatial multiplexing (CLSM) is active. With the inclusion of CLSM in JT CoMP, the obtained momentary channel state information can be utilized by coordinated eNBs for appropriate network gain enhancement. The simulation results demonstrate significant improvement in mean throughput and cell edge throughput for a CoMP based HetNet compared to a non-CoMP based HetNet with respect to UE velocity. The effectiveness of CLSM is found to degrade as UE velocity increases which is expected due to poor feedback capabilities of high velocity UEs. In contrast, simulation results show that the CLSM integrated inter-site based JT CoMP network provides improved reception for high velocity, while the intrasite-based CoMP network delivers better service at lower velocities.
{"title":"Joint Transmission Coordinated Multipoint on Mobile Users in 5G Heterogeneous Network","authors":"Tahmina Khanom Tandra, Fehima Tajrian, Afia Hossain, M. T. Kawser, Mohammad Rubbyat Akram, A. Shams","doi":"10.1109/CITDS54976.2022.9914149","DOIUrl":"https://doi.org/10.1109/CITDS54976.2022.9914149","url":null,"abstract":"To broaden user experience and support a wide range of bandwidth hungry devices cellular operators are adopting 5G network. However, the predominance of Inter-Cell Interference (ICI) in 5G stands as a hurdle for cell edge UEs. With increase in UE velocity, the effect of Doppler shift becomes more prominent, resulting in a significant drop in cell edge and mean data rates. A potential solution to improve the network service quality in the cell edge and reduce the impact of ICI for mobile users is to provide the UE with the best signal quality through coordination among multiple eNodeBs (eNB) located in different cell sites i.e., by virtually forming a massive antenna array with the coordinated eNB, a technique popularly known as Joint Transmission Coordinated Multipoint (JT CoMP). This paper investigates the performance of JT CoMP based heterogeneous network (HetNet) for UEs at different velocities while closed loop spatial multiplexing (CLSM) is active. With the inclusion of CLSM in JT CoMP, the obtained momentary channel state information can be utilized by coordinated eNBs for appropriate network gain enhancement. The simulation results demonstrate significant improvement in mean throughput and cell edge throughput for a CoMP based HetNet compared to a non-CoMP based HetNet with respect to UE velocity. The effectiveness of CLSM is found to degrade as UE velocity increases which is expected due to poor feedback capabilities of high velocity UEs. In contrast, simulation results show that the CLSM integrated inter-site based JT CoMP network provides improved reception for high velocity, while the intrasite-based CoMP network delivers better service at lower velocities.","PeriodicalId":271992,"journal":{"name":"2022 IEEE 2nd Conference on Information Technology and Data Science (CITDS)","volume":"20 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114722533","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-16DOI: 10.1109/CITDS54976.2022.9914012
Anca Alexan, A. Alexan, S. Oniga
Recognition of activities today is an essential element in the artificial intelligence field. The smart environment is more and more present in residential spaces, which makes the activities recognition algorithms more and more efficient and adaptable. This article addresses the issue of recognizing activity based on event density. Each activity is interpreted as a density graph, and the recognition is done using image processing algorithms. This method facilitates the determination of transition zones between activities. Each user performs each activity differently, which makes it difficult to recognize the activities. The activities abstracting method presented in this article improves the recognition rate of activities.
{"title":"Single user activity recognition with Density Activity Abstraction Graphics Algorithm","authors":"Anca Alexan, A. Alexan, S. Oniga","doi":"10.1109/CITDS54976.2022.9914012","DOIUrl":"https://doi.org/10.1109/CITDS54976.2022.9914012","url":null,"abstract":"Recognition of activities today is an essential element in the artificial intelligence field. The smart environment is more and more present in residential spaces, which makes the activities recognition algorithms more and more efficient and adaptable. This article addresses the issue of recognizing activity based on event density. Each activity is interpreted as a density graph, and the recognition is done using image processing algorithms. This method facilitates the determination of transition zones between activities. Each user performs each activity differently, which makes it difficult to recognize the activities. The activities abstracting method presented in this article improves the recognition rate of activities.","PeriodicalId":271992,"journal":{"name":"2022 IEEE 2nd Conference on Information Technology and Data Science (CITDS)","volume":"45 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129298272","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-16DOI: 10.1109/CITDS54976.2022.9914268
Yuping Yan, P. Ligeti
Federated learning (FL) provides a higher privacy guarantee for data sharing in a multi-party computation environment. However, how to invite participants to federated training if they already have a self-sanitized dataset? What is more, FL can not be directly applied to Non-IID data, and the global model can not meet the different feature requirements of clients. Personalized and incentive mechanisms are very necessary to build a good learning environment for FL. However, there has been little discussion about personalized and incentive mechanisms schemes so far, while more attention is focused on the optimization, efficiency and effectiveness improvement, and security aspects. Thus, in this paper, we make a review of personalized and incentive mechanisms of federated learning with different techniques.
{"title":"A Survey of Personalized and Incentive Mechanisms for Federated Learning","authors":"Yuping Yan, P. Ligeti","doi":"10.1109/CITDS54976.2022.9914268","DOIUrl":"https://doi.org/10.1109/CITDS54976.2022.9914268","url":null,"abstract":"Federated learning (FL) provides a higher privacy guarantee for data sharing in a multi-party computation environment. However, how to invite participants to federated training if they already have a self-sanitized dataset? What is more, FL can not be directly applied to Non-IID data, and the global model can not meet the different feature requirements of clients. Personalized and incentive mechanisms are very necessary to build a good learning environment for FL. However, there has been little discussion about personalized and incentive mechanisms schemes so far, while more attention is focused on the optimization, efficiency and effectiveness improvement, and security aspects. Thus, in this paper, we make a review of personalized and incentive mechanisms of federated learning with different techniques.","PeriodicalId":271992,"journal":{"name":"2022 IEEE 2nd Conference on Information Technology and Data Science (CITDS)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128167417","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-16DOI: 10.1109/CITDS54976.2022.9914096
A. Rumyantsev, R. Nekrasova, S. Astafiev, A. Golovin
In this paper we apply regenerative simulation and distributed computing to study the energy efficiency of a supercomputer with speed scaling. We use generalized semi-Markov processes to simulate the supercomputer in steady state, and perform exhaustive search of the optimal speed scaling policy in a small-scale heterogeneous model where the per-class amount of work has a heavy-tailed distribution. The preliminary simulation results are reported, which demonstrate the capabilities of the software packages used.
{"title":"Distributed Regenerative Simulation of a Speed Scaling Supercomputer","authors":"A. Rumyantsev, R. Nekrasova, S. Astafiev, A. Golovin","doi":"10.1109/CITDS54976.2022.9914096","DOIUrl":"https://doi.org/10.1109/CITDS54976.2022.9914096","url":null,"abstract":"In this paper we apply regenerative simulation and distributed computing to study the energy efficiency of a supercomputer with speed scaling. We use generalized semi-Markov processes to simulate the supercomputer in steady state, and perform exhaustive search of the optimal speed scaling policy in a small-scale heterogeneous model where the per-class amount of work has a heavy-tailed distribution. The preliminary simulation results are reported, which demonstrate the capabilities of the software packages used.","PeriodicalId":271992,"journal":{"name":"2022 IEEE 2nd Conference on Information Technology and Data Science (CITDS)","volume":"29 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130412134","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：